Real-time EEG denoising – LabView/FPGA implementation

The accurate recording of EEG (electro-encephalographic) signals from the scalp provides an important source of information for many research topics focused on analyzing the correlation between brain dynamics and specific states of the body for patients or healthy subjects. The recording of clean EEG signals becomes nevertheless a daunting task when attempting to quantify the neural response to galvanic vestibular stimulation (GVS). In such a case, about 90% of the signal is due to the electric signal propagation from the points of current application to the EEG electrodes, and consequently the collected data has a very low signal to noise ratio.  The modeling of this causal dependence is important for designing biofeedback mechanisms, and therefore requires a real-time signal processing for significantly reducing the noise components.

The proposed project targets the implementation of a real-time denoising algorithm for EEG responses to GVS. The signals are collected with the help of an EEG electrode array placed on a special helmet and amplified by the signal conditioning circuitry associated with each of the electrodes. The data stream delivered by the commercial EEG helmet will be interfaced with an FPGA-based intelligent data acquisition module that can be programmed directly in LabView. A dedicated adaptive algorithm will be firstly implemented in LabView/Matlab for off-line denoising of the recorded EEG data, while in a second phase the algorithm will be mapped to a dedicated hardware defined in FPGA, for real-time processing.

The candidate will have the chance to combine theoretical aspects of biomedical signal processing with practical hardware implementation in the reconfigurable module and to assist the experimental data gathering process in the Sensorimotor Physiology Laboratory.

Intern: 
Nakul Sharma
Faculty Supervisor: 
Dr. Edmond Cretu
Province: 
British Columbia
Sector: